nature immunology • volume 4 no 2 • february 2003 • www.nature.com/natureimmunology
ARTICLES
124
I-hsin Su1,Ashwin Basavaraj1,Andrew N. Krutchinsky2, Oliver Hobert3,Axel Ullrich4,Brian T. Chait2 and Alexander Tarakhovsky1
Published online 23 December 2002; doi:10.1038/ni876
Polycomb group protein Ezh2 is an essential epigenetic regulator of embryonic development in mice,but its role in the adult organism is unknown. High expression of Ezh2 in developing murinelymphocytes suggests Ezh2 involvement in lymphopoiesis. Using Cre-mediated conditional muta-genesis, we demonstrated a critical role for Ezh2 in early B cell development and rearrangement ofthe immunoglobulin heavy chain gene (Igh). We also revealed Ezh2 as a key regulator of histone H3methylation in early B cell progenitors. Our data suggest Ezh2-dependent histone H3 methylation as anovel regulatory mechanism controlling Igh rearrangement during early murine B cell development.
1Laboratory of Lymphocyte Signaling and 2Laboratory of Mass Spectrometry and Gaseous Ion Chemistry,The Rockefeller University, New York, NY 10021 USA. 3Department ofBiochemistry and Molecular Biophysics, Center for Neurobiology and Behavior, Columbia University, College of Physicians and Surgeons, New York, NY 10032, USA. 4Max Planck
Institute for Biochemistry, Department of Molecular Biology, D-82152 Martinsried, Germany. Correspondence should be addressed to A.T. ([email protected]).
Ezh2 controls B cell developmentthrough histone H3 methylation and
Igh rearrangement
Commitment of cells to a particular lineage and their maintenance in a dif-ferentiated state involves the activation of limited gene sets while leavingthe rest of the genome in a repressed state1. The cell type–specific geneexpression pattern is stabilized by changes in the chromatin structure asso-ciated with active and silent genomic loci2. These heritable chromatin mod-ifications can be maintained by counteraction of transcriptional activatorsof the trithorax group (TrxG) proteins and repressors of the polycombgroup (PcG) proteins2.
The PcG proteins are organized in two to five megadalton (MD) com-plexes that exert their influence on gene expression through local chro-matin modifications3. At least two distinct mechanisms of gene repressionare utilized by PcG proteins. The PcG complex that contains the EED,EZH2 and EZH1 proteins can control gene repression through the recruit-ment of a histone deacetylase followed by local chromatin deacetylation4,5.The other PcG complex—polycomb repressive complex 1 (PRC-1)—con-tains polycomb 2 (HPC2), polyhomeotic (HPH), BMI1 and Ring-fingerprotein 1 (RING1) proteins that negatively regulate chromatin accessibili-ty promoted by the chromatin remodeling SWI-SNF complex6.
The physiological significance of PcG protein–mediated gene regulationis underscored by developmental abnormalities found in mice deficient inthe PcG proteins. Targeted disruption of the gene encoding PcG proteinresults in ectopic expression of Hox transcription factors, which controlcell type–specific gene expression7. As a consequence, mice deficient fordistinct PcG proteins develop skeletal transformations8, show male-to-female sex reversal9 or neurological abnormalities8.
A common consequence of PcG deficiencies is the defective develop-ment and activation of lymphocytes. For example, inactivation of mam-malian homologs of the Drosophila gene posterior sex combs (Psc), Bmi1or mel-18 (also termed Zfp144), causes a severe block in B cell develop-ment that leads to B cell lymphopenia in the mutant mice8,10. Similar toBmi1 and mel-18, the presence of rae28, a mammalian ortholog of the
Drosophila gene polyhomeotic (Ph), is required for normal B cell develop-ment, as shown by reduced generation of pre-B and immature B cells fromrae28-deficient fetal liver hematopoietic progenitors11. Deficiency in Cbx2(also termed M33), the murine counterpart of the Drosophila gene poly-comb (Pc), does not affect lymphocyte development but renders splenic Bcells unresponsive to lipopolysaccharide (LPS)12.
Several lines of evidence support the involvement of the PcG proteinEzh2—a mammalian homolog of the Drosophila PcG protein E(Z) (origi-nally termed Enx-1)13,14—in chromatin remodeling, as well as in lympho-cyte development and activation. Ezh2 is distinct among the PcG family ofgenes because it contains the evolutionarily conserved SET domain that isresponsible for histone H3 methyltransferase activity (HMTase) of E(Z).The E(Z) methylates in vitro lysine 9 (H3-K9) and lysine 27 (H3-K27) ofhistone H315–17. In general, methylation of lysines within the histone H3 N-terminal tail causes stable changes in chromatin that define the activationstatus of the gene. Thus, the methylation of H3-K9 or H3-K4 by distinctHMTases is associated with stable gene repression or transcriptional acti-vation, respectively18. However, neither the ability of mouse Ezh2 to con-trol histone H3 methylation in vitro and in vivo nor the physiological sig-nificance of Ezh2-mediated H3 lysine methylation are known.
In mice, Ezh2 is most abundant at sites of embryonic lymphopoiesis,such as fetal liver and thymus14. Up-regulation of Ezh2 in proliferatinghuman germinal center B cells (centroblasts)19 and mitogen-stimulatedlymphocytes20 suggests an important role for this protein in B cell division.The potential importance of Ezh2 in lymphocyte activation is further sup-ported by its association with Vav, one of the key regulators of the recep-tor-mediated signaling in lymphocytes13.
Functional analysis of Ezh2 in lymphocyte development is complicatedby the early embryonic death of Ezh2-deficient mice21. To circumvent thelethal effect of the Ezh2 null mutation, we employed the Cre-loxP technol-ogy for conditional gene inactivation22. Here we demonstrate that Ezh2
www.nature.com/natureimmunology • february 2003 • volume 4 no 2 • nature immunology 125
controls B cell development through the regulation of histone H3 methyla-tion and immunoglobulin heavy chain gene (Igh) rearrangement. We sug-gest that Ezh2-dependent histone H3 methylation leads to chromatin mod-ification required for normal Igh rearrangement, which is critical for earlyB cell development.
ResultsEzh2 expression in B lineage cellsAnalysis of Ezh2 mRNA expression in B lineage cells isolated from wild-type 129Sv mice showed a reverse correlation between Ezh2 expressionand the degree of B cell differentiation and maturation (Fig. 1). In B cellprogenitors, Ezh2 expression was highest in pro-B cells (IgM–B220+
CD43+) and was much lower in pre-B cells (IgM–B220+CD43–). Furtherdifferentiation of B cells led to a reduction in Ezh2 expression in imma-ture B cells (B220intIgM+gD–) and, most profoundly, in mature recirculat-ing (B220hiIgM+IgD+) B cells. In contrast to Ezh2, the expression levels ofEzh1, which harbors a SET domain and shows 67% amino acid identityto Ezh223, was low in pro-B and pre-B cells but increased in immature andrecirculating B cells. The high abundance of Ezh2 mRNA in early B cell
progenitors points to a possible involvement of Ezh2 in the regulation ofearly B cell development in the bone marrow, whereas Ezh1 may play arole in the peripheral B cells.
Conditional inactivation of Ezh2Using a gene targeting approach, we generated embryonic stem (ES) cellsand mice in which exons encoding the SET domain of Ezh2 are flankedwith loxP sequences, which can be recognized by Cre recombinase (Fig.2a,b). The SET domain was originally described as a sequence homolog tothe three Drosophila genes suppressor of variegation (3–9) (Su(var)3-9),enhancer of zeste (E(Z)) and trithorax (trx)24. The SET domain was chosenas a target for Ezh2 inactivation because this domain is essential for func-tioning of the Drosophila homolog of Ezh2, E(Z)25.
Insertion of the two loxP sequences into the Ezh2 genomic locus did notalter Ezh2 expression, and mice homozygous for the loxP-flanked Ezh2allele (Ezh2fl/fl) were viable and developed normally. By crossing Ezh2fl/fl
mice with transgenic mice that expressed Cre recombinase in the germ line,we generated mice heterozygous for the modified Ezh2 allele (Ezh2–/+).Whereas Ezh2–/+ mice developed and lived normally, crossing of Ezh2–/+
mice never yielded any Ezh2–/– pups. We took the fact that the deletion ofexons encoding the SET domain in Ezh2 led to embryonic lethality asproof of the reliability of the chosen strategy of Ezh2 inactivation, as com-plete deletion of Ezh2 in the germ line also resulted in early death of themutant embryos21.
To achieve the inducible inactivation of Ezh2, Ezh2fl/+ mice were bred toMx-Cre transgenic mice harboring the Cre-recombinase transgene drivenby the interferon-inducible Mx promoter26. The deletion of Ezh2 wasinduced by repetitive intraperitoneal (i.p.) injections of poly(I)•poly(C)26.Unless specified, mice received three poly(I)•poly(C) injections at 2-dayintervals and were sacrificed on day 10 following the last injection.Southern blot analysis of the DNA derived from various lymphoid organsof the poly(I)•poly(C)-injected Mx-Cre Ezh2fl/fl mice revealed virtually
Figure 1.Expression of Ezh2 and Ezh1 mRNA in B lineage cells.Total RNA iso-lated from FACS-purified B cell subpopulations was reverse-transcribed with oligo(dT)primers and threefold serial dilutions of cDNA were used for PCR amplification.ThePCR products were visualized by Southern blotting with Ezh2- or Ezh1-specific probes.The two Ezh2-specific PCR products reflect alternative splicing of the Ezh2 mRNA. RT-PCR analysis of HPRT mRNA expression was used as a cDNA loading control.
Figure 2. Conditional inactivation of Ezh2. (a) The domain structure of Ezh2 protein, part of the Ezh2 genomic locus and the Ezh2-targeted locus before and after Cre-mediated recombination are shown.The small open boxes indicate exons encoding the SET domain.The large open box corresponds to the last exon of Ezh2 encoding the3′ untranslated region.The arrows correspond to the loxP sequences. neor, neomycin phosphotransferase gene.The sizes of the expected restriction enzyme–digested DNAfragments that were recognized by probe A (the 1-kb EcoRI-ManI fragment) are shown. (b) Genomic DNA isolated from ES cells was digested with BamHI and analyzed bySouthern blotting. Probe A recognizes a 19-kb fragment derived from the wild-type allele (lanes 1 and 2) and a 24.5-kb fragment from the neor-containing targeted allele (lane1). DNA isolated from the pIC-Cre transiently transfected ES clones was digested with KpnI-BamHI. Loss of neor only or together with the loxP-flanked exons gave rise to 8.5-kb (lane 4) and 5-kb (lanes 3 and 4) DNA fragments, respectively.To distinguish between the wild-type or Cre-mediated deleted Ezh2 alleles, DNA was digested with EcoRI.The 4.5-kb and 7.2-kb DNA fragments correspond to the deleted (lane 3′) or loxP-flanked (lanes 3′ and 4′) alleles, respectively. (c) Ezh2fl/lf (lane 1) and Ezh2fl/fl Mx-Cre (lane 2)mice were injected with poly(I)•poly(C).The efficiency of the Cre-mediated Ezh2 deletion was quantified by densitometry analysis of Southern blots hybridized with probe A.The loxP-flanked (Ezh2 fl) and deleted (Ezh2 –) Ezh2 alleles gave rise to 7.2-kb and 4.5 kb DNA fragments, respectively. The percentage values indicate the deletion efficiency.
nature immunology • volume 4 no 2 • february 2003 • www.nature.com/natureimmunology
ARTICLES
126
complete inactivation of Ezh2 in the bone marrow (98%) and thymus(95%); however, in the spleen, deletion occurred with 50% efficiency (Fig.2c). Although cells homozygous for the Ezh2 deletion contained mRNAcorresponding to the Cre-modified Ezh2 gene (Ezh2–/–), SET domain–defi-cient truncated Ezh2 protein could not be detected (data not shown). Thesedata indicated that deletion of exons encoding the SET domain leads tocomplete Ezh2 inactivation.
Impaired development of Ezh2-deficient B cellsThe frequencies and absolute numbers of cells comprising various lym-phocyte subpopulations were similar in the bone marrow and peripherallymphoid organs of the poly(I)•poly(C)-injected Ezh2fl/fl and PBS-inject-ed Mx-Cre Ezh2fl/fl mice analyzed on day 10 following the last injection(Fig. 3a). This result excluded the toxicity of poly(I)•poly(C) as a possi-ble cause of changes in B cell development. Analysis of Ezh2-deficientbone marrow cells derived from the poly(I)•poly(C)-injected Mx-CreEzh2fl/fl mice showed unaltered development of pro-B cells(IgM–B220+CD43+) followed by a decrease in the frequencies and num-bers of pre-B (IgM–B220+CD43–) and immature B cells (IgM+B220int).Complete deletion of Ezh2 in pro-B cells was confirmed by Southernblot, polymerase chain reaction (PCR) and reverse-transcribed PCR (RT-PCR) analysis (Supplementary Fig. 1 online). Collectively, the observedalteration in development of Ezh2-deficient B cells revealed Ezh2 as acomponent of the checkpoint mechanism that controls the pro-B to pre-B cell transition.
Deletion of Ezh2 at nearly 100% efficiency in the bone marrow of thepoly(I)•poly(C)-injected Mx-Cre Ezh2fl/fl mice probably also led to Ezh2-deficiency in stromal cells that support B cell development via the secre-tion of various cytokines27. Hence, observed changes in the development ofEzh2-deficient B cells may not have been cell-autonomous. To address thisquestion, bone marrow transfer experiments were performed. To distin-guish the donor bone marrow–derived cells from the recipient cells, surfaceexpression of Ly9.1 was used as a marker for the donor-derived cells.Ly9.1+ Ezh2-deficient bone marrow cells were isolated frompoly(I)•poly(C)-treated Mx-Cre Ezh2fl/fl mice and transferred either aloneor in combination with the wild-type bone marrow cells into lethally irra-diated Ly9.1-negative C57BL/6 mice. The resulting bone marrow chimeraswere analyzed at different time points after transplantation. The analysis ofthe bone marrow chimeras 28 days after transplantation revealed a block inpro-B to pre-B cell development, similar to that observed in the bone mar-row of poly(I)•poly(C)-injected Mx-Cre Ezh2fl/fl mice (SupplementaryFig. 2 online). Cotransfer of the wild-type and Ezh2–/– bone marrow cellsdid not revert the block in Ezh2–/– B cell development and the block was soprofound that even twenty weeks after transplantation, the number ofEzh2–/– IgM+ B cells was below 1% of the control values (data not shown).
These data prove the cell-autonomous nature of defective Ezh2-deficient Bcell development.
Reduced µ chain expression in Ezh2-deficient cellsThe pro-B to pre-B cell differentiation and expansion of pre-B cells aregoverned by signals derived from the surface expressed pre-B cell receptor(pre-BCR)28. Thus defects either in pre-BCR formation or impairment ofits signaling properties may severely impair the pro-B to pre-B cell devel-opment. About 25–30% of the pro-B cells derived from control miceexpressed the intracellular µ chain, as determined by intracellular stainingwith the µ chain–specific antibody M41. In contrast to the control pro-Bcells, the Ezh2-deficient pro-B cells did not form a well-defined populationof B220int, intracellular µ+ cells and the overall frequency of B220+, intra-cellular µ+ cells was reduced to 30% (Fig. 3b). The reduction in frequen-cies of µ chain expressing cells correlates directly with a three-fold reduc-tion in µ chain mRNA expression in Ezh2–/– pro-B cells compared to con-trol pro-B cells. In contrast, Ezh2-deficiency did not affect the frequency ofpro-B cells expressing intracellular κ light chain (Supplementary Fig. 3online). The latter result therefore suggests Ezh2 specifically affects µchain production.
Igh transgene rescues Ezh2-deficient B cell developmentIf the defective heavy chain rearrangement is the main cause of impairedEzh2-deficient B cell development, the expression of a prerearrangedheavy chain should rescue Ezh2-deficient pro-B cells. To test such a possi-bility, Mx-Cre Ezh2fl/fl mice were bred with mice carrying rearranged VHD-HJH (B1-8) inserted at the JH locus (B1-8i) by homologous recombination29.The Ezh2fl/fl Mx-Cre B1-8i mice were injected with poly(I)•poly(C) and theB cell population was analyzed on day 10 after the last injection. At thattime, deletion of Ezh2 in the bone marrow was 100%. Expression of B1-8iheavy chain rescued B cell development in the bone marrow (Fig. 4a).
The low efficiency of the poly(I)•poly(C)-induced Ezh2 inactivation inthe peripheral B cells of Mx-Cre Ezh2fl/fl B1-8i mice (50%) precluded thepossibility of analysis of B cell function in these mice. Nonetheless, highefficiency of poly(I)•poly(C)-induced Ezh2 inactivation in the bone mar-row cells of Ezh2fl/fl Mx-Cre B1-8i mice allowed us to generate bone mar-row chimeras carrying exclusively B1-8i–expressing Ezh2-deficientperipheral B cells. Transfer of bone marrow cells derived frompoly(I)•poly(C)-injected Ezh2fl/fl Mx-Cre B1-8i mice into sublethally irra-diated recombination-activating gene 1–deficient (RAG-1–/–) mice result-ed in the generation of subpopulations of developing bone marrow andperipheral B cells in the recipient mice that were similar to those observedin the donor mice (Fig. 4b,c). Ezh2-deficient peripheral B cells maturednormally (Fig. 4c). Moreover, deficiency of Ezh2 did not affect Ig switchrecombination in vitro, as shown by the wild-type frequencies of the
a b Figure 3. Impaired development ofEzh2-deficient B cells. (a) Bone mar-row cells were isolated from the controland experimental mice on day 10 afterthe last PBS or poly(I)•poly(C) injection.Expression of the indicated surface pro-teins was analyzed by FACS. The lowerpanel shows the expression of CD43and B220 within the subpopulation ofsurface IgM–negative (sIgM–) bone mar-row cells. (b) The expression of intracel-lular µ chain in B220+CD43+ pro-B cellswas analyzed by FACS. The numbersindicate the percentages of gated cells.The FACS data are representative of tenindependent experiments.
www.nature.com/natureimmunology • february 2003 • volume 4 no 2 • nature immunology 127
IgG1-positive cells induced by splenic B cell incubation with 10 µg/ml ofLPS and 25 U/ml of interleukin 4 (IL-4) (Fig. 4d). This result suggeststhat lack of µ chain expression but not defective pre-BCR signaling is achief cause of impaired B cell development in the absence of Ezh2.
Ezh2 is dispensable for peripheral B cellsExpression of the transgenic heavy chain may potentially mask changes inmaturation and activation of peripheral Ezh2-deficient B cells. To achieveEzh2 inactivation in peripheral B cells expressing a wild-type BCR reper-toire, the Ezh2fl/fl mice were bred to mice expressing the Cre recombinasegene under control of the Cd19 promoter (CD19-Cre)30. The analysis ofEzh2 mRNA expression in subpopulations of developing and peripheral Bcells showed the presence of mRNAs corresponding to the wild-type anddeleted Ezh2 alleles in pro-B cells. Further B cell development was accom-panied by loss of the wild-type Ezh2 mRNA expression (SupplementaryFig. 4 online). Complete (100%) Ezh2 inactivation in peripheral B cellswas also confirmed by Southern blot analysis (data not shown). Theperipheral Ezh2-deficient B cells developed normally, as defined by mem-brane-bound expression of IgM and IgD (Supplementary Fig. 4 online).Moreover in vitro proliferation of splenic B cells in response to anti–IgMF(ab)2, anti-CD40 or LPS was not affected in the absence of Ezh2(Supplementary Fig. 4 online). Also Ezh2 deficiency had no impact onearly BCR-mediated signaling, as defined by the wild-type–like kinetics ofanti-IgM–induced calcium mobilization (Supplementary Fig. 4 online).Thus, Ezh2 appears dispensable for the maturation and activation ofperipheral B cells.
Ezh2 regulates VHJ558 gene rearrangementNext, we attempted to address the mechanism responsible for impaired µchain expression in the absence of Ezh2. Reduced µ chain mRNA expres-sion may result from inefficient V(D)J rearrangement, lower transcriptionof the rearranged Igh genes or decreased stability of the transcribedmRNA. The first possibility was addressed by the comparative analysis ofIgh rearrangement in wild-type and Ezh2-deficient pro-B cells. Therearrangement of two different VH gene families, VH7183 and VHJ558, wasexamined. The most fundamental difference between these families lies intheir proximity to the DJH element. The VH7183 family that consists ofaround 25 VH genes is the smallest VH gene family and is adjacent to theDH segments at the most 3′ end of VH gene locus31. Contrary to VH7183,the VHJ558 family is the largest and most frequently rearranged VH gene
family that occupies the most 5′ end to the middle of the VH locus31,32. TwoVH gene primers, which are specific for most of the VH gene segmentswithin VHJ558 family or VH7183 family, respectively33 were used to ampli-fy the rearranged VH to DJH1 or DJH2 sequences. The incidence of VH toDJH joints involving VH7183 segments was similar in control and Ezh2–/–
pro-B cells (Fig. 5a). In contrast, recombination of VHJ558 segments wasreduced to 25% of the control (Fig. 5a). As a consequence, the amount ofµ chain transcripts corresponding to VHJ558 in pro-B cells was reduced to11% of the wild-type, whereas µ chain transcripts expression correspond-ing to the VH81X gene, the most prevalent gene within VH7183 family,remained unaffected (Fig. 5b).
The discrepancy between 25% reduced VHJ558 rearrangement and11% reduced VHJ558 mRNA levels could be explained by the presence ofa substantial fraction of nonfunctional VHJ558 rearrangements. Ig tran-scripts encoded by nonproductively rearranged VHDHJH are unstable34. Totest whether the reduction of µ chain VHJ558 transcript in Ezh2-deficientpro-B cells reflected an increased frequency of nonfunctional VHDHJH
rearrangement, the individual VHDHJH joints derived from the Ezh2–/– orcontrol pro-B cells (B220+CD43+HSA+BP-1+) were amplified by PCR,sequenced and compared to the published VH gene sequences.Productively rearranged VHDHJH sequences were present in 60% of con-trol and 20% of Ezh2–/– pro-B cells. In addition, the repertoire of VHJ558rearrangement was more diverse in control pro-B cells, whereas in Ezh2–/–
pro-B cells, the VH gene usage was restricted to a limited number of VH
gene segments (Fig. 5c). There was no major difference between controland Ezh2–/– pro-B cells in the percentage of productive V(D)J rearrange-ments and VH gene usage within theVH7183 family (Fig. 5c). This resultshows the selective involvement of Ezh2 in VHJ558 gene rearrangement.
Ezh2 does not control Igh germline transcriptionDiminished rearrangement of VHJ558 genes in Ezh2-deficient pro-B cellscould be due to poor accessibility of this particular VH locus to the recom-bination machinery. The degree of accessibility of the Igh locus correlateswith transcription of the Igh genes in germline configuration (germlinetranscripts)35. The expression levels of different germline transcripts wereanalyzed in control and Ezh2–/– pro-B cells. The levels of germline tran-scripts of VHJ558 were even higher in Ezh2–/– pro-B cells compared to con-trol cells (Fig. 6a). The germline transcripts corresponding to proximal VH
genes such as VH81X were equally abundant in control and Ezh2–/– pro-Bcells. The expression levels of Iµ and µ0 germline transcripts in control and
a b
c dFigure 4. Expression of transgenic BCR rescues the devel-opment of Ezh2-deficient B lineage cells. Bone marrow cellsisolated from poly(I)•poly(C)-treated mice were either analyzedimmediately (a) or transferred into sublethally irradiated RAG-1–/–
C57BL/6 mice (b,c).The frequencies of developing sIgM– pro-B andpre-B cell subpopulations in the bone marrow (a,b) or splenic Bcells (c) were analyzed by FACS. Numbers indicate the percentagesof gated cells. The data are representative of three independentexperiments. (d) Purified splenic B cells were cultured in vitro in thepresence of LPS and IL-4 and analyzed 3 days after stimulation byFACS. Cells treated with LPS only were used as a negative control.
nature immunology • volume 4 no 2 • february 2003 • www.nature.com/natureimmunology
ARTICLES
128
Ezh2–/– pro-B cells were similar (Fig. 6a). The observed dichotomybetween the levels of VHJ558 genes germline transcription and VHJ558genes rearrangement suggest that another mechanism, unrelated to tran-scription, controls VHJ558 gene rearrangement.
Ezh2-deficient cells are equipped for rearrangementThe expression levels of RAG-2, DNA-PK and Ku80 mRNAs, which areessential for the V(D)J recombination36, were unaltered in Ezh2-deficientpro-B cells (Supplementary Fig. 5 online). Furthermore, the analysis ofthe DNA double-stranded breaks (DSBs) that occur during the course of
VHJ558 to DJH rearrangement35 did not reveal an accumulation of the sig-nal and coding ends in Ezh2-deficient pro-B cells (Supplementary Fig. 5online). These results suggest that Ezh2-deficient pro-B cells are wellequipped for V(D)J rearrangement, and those recombination events that areinitiated should be completed.
Ezh2-deficient pro-B cells live and divide normallyIn B cell ontogeny, the proximal VH7183 genes are preferentiallyrearranged in the pro-B cells32, whereas B cells possessing VHJ558 heavychains expand at later stages of B cell development37. If Ezh2 controls
Figure 5. Impaired rearrangement and expression of VHJ558 family genes in Ezh2-deficient pro-B cells. (a) Primers specific for VH genes and primers recognizing thesequence 3′ downstream of JH2 were used to amplify rearranged VH to DJH1 (780-bp) and DJH2 (470-bp) from genomic DNA isolated from pro-B cells. DNA isolated from JHT micewas used as a negative control.RT-PCR analysis of Thy1.2 mRNA expression was used as a cDNA loading control. (b) Degenerate primers that recognize most of the VH genes (MsVHE)and a µ constant region–specific primer (MsCµE) were used to amplify total µ chain transcript. µ chain transcripts corresponding to the VH81X gene or VHJ558 family were amplifiedwith a specific set of primers (Supplementary Table 1 online).The PCR products were visualized by Southern blotting with an IgH constant region–specific probe (MsCµN). RT-PCR analysis of Igβ mRNA expression was used as a cDNA loading control. (c) VHJ558 and VH7183 were amplified by PCR with DNA isolated from pro-B cells.The individual VHD-HJH regions were sequence-analyzed with the DNAPLOT Program Package and Medline BLAST search.The VH gene names or sequence accession numbers are indicated
a
b
c
a
b
c
Figure 6. Ezh2 does not control expression of IgH germline transcripts, IL-7–mediatedSTAT5 activation or histone acetylation in pro-B cells. (a) Expression levels of Iµ, µ0 and VH
germline transcripts in pro-B cells were analyzed by RT-PCR and Southern blotting with µ constantregion– (Iµ, µ0) or VH–specific probes. RT-PCR analysis of Igβ mRNA expression was used as a cDNA load-ing control. (b) Thymocytes or pro-B cells were incubated in the presence or absence of IL-7 (20 ng/ml)and STAT5 activation was analyzed by EMSA. (c) ChIP was done on pro-B cells with anti-acetyl–histoneH3, anti-acetyl–histone H4 or without antibody (negative control).The coprecipitated DNA was analyzedby PCR with VH gene–specific primers or primers recognizing the Eµ enhancer region, then the PCR prod-ucts were analyzed by Southern blotting with specific probes.
www.nature.com/natureimmunology • february 2003 • volume 4 no 2 • nature immunology 129
pro-B cell survival, the observed changes in VHJ558 rearrangement maysimply reflect the premature death of Ezh2-deficient pro-B cells prior tocompletion of VHJ558 rearrangement. However the viability of pro-Bcells, as defined by terminal deoxynucleotidyltransferase–mediateddUTP-biotin nick end–labeling (TUNEL) assay or intracellular stainingof activated caspases, was not affected by Ezh2 deficiency(Supplementary Fig. 6 online).
The regulation of V(D)J recombination is coupled to the cell cycle38. TheRAG-1 and RAG-2 proteins are abundant during the G1 phase and V(D)Jrecombination is prohibited during S and M phase38. Hence changes in thecell cycle progression could possibly affect the Igh rearrangement.However, unaltered cell cycle parameters of Ezh2-deficient pro-B cellsshown by in vivo 5-bromodeoxyuridine (BrdU) labeling argued againstsuch possibility. (Supplementary Fig. 6 online). Thus, the Igh rearrange-ment defect in Ezh2-deficient pro-B cells was not due to poor survival ofthe cells or defective cell cycle progression.
Unimpaired STAT5 activation and histone acetylationIgh rearrangement in general and histone acetylation of the VHJ558locus is regulated by IL-739,40. However, IL-7Rα expression and the
IL-7–induced activation of STAT541 were not reduced in Ezh2–/–
CD19+IgM– B cell progenitors (Fig. 6b and Supplementary Fig. 7online). Moreover, the acetylation of histones H3 and H4 associatedwith Igh locus was not impaired by Ezh2 deficiency (Fig. 6c). Thisresult shows that Ezh2 does not control IL-7 signaling upstream ofSTAT5 or signaling that leads to histone H3 and H4 acetylation.
Ezh2 methylates histone H3 in pro-B cellsThe SET domain of Ezh2 may possess HMTase activity similar to E(Z),which could result in impaired histone H3 methylation in Ezh2-deficientcells. Immunoblot analysis of histone H3 methylation with pan–methyl-lysine–specific antibody revealed reduced histone H3 lysine methylationin the Ezh2-deficient pro-B cells (62% reduction), compared to controlpro-B cells (Fig. 7a). Incubation of control pro-B cells with IL-7 (20ng/ml) caused a modest but consistent increase of lysine methylation, asdetermined by immunoblotting and fluorescence-activated cell sorting(FACS) (Fig. 7a–c and Supplementary Fig. 8 online). In contrast, incu-bation of Ezh2-deficient cells with IL-7 did not increase histone H3lysine methylation (Fig. 7a–c). Methylation was unaltered at H3 lysine atposition 4 (H3-K4) and a modest reduction of H3 lysine 9 (H3-K9)
Figure 7. Reduced lysine methylation of histone H3 in Ezh2-deficient pro-B cells. Methylation of histone H3 innuclear lysates of pro-B cells incubated with or without IL-7 was analyzed by immunoblotting with anti-pan–methyl-lysine(a,b), anti-dimethyl–histone H3-K9 or anti-dimethyl– histone H3-K4 (a).The amount of histone H3 loading was controlledfor by immunoblotting with anti–histone H3.The numbers indicate fold changes compared to the signal of the unstimu-lated control lysate once it had been normalized against the amount of histone H3 (a). Fold changes from four indepen-dent experiments are summarized in b. Bars represent mean ± s.d. data.The differences between control pro-B cells thatdidn’t or did receive IL-7 stimulation were statistically significant and are indicated by asterisks (Student’s t-test,P < 0.005).(c) Kinetics of IL-7–induced methylation.The levels of intracellular lysine methylation were analyzed by FACS.The median fluorescence intensity of methylated lysine staining at eachtime point is shown. (d) Portion of the mass spectra of tryptic digests of histone H3 comparing methylation of the peptide 27KSAPATGGVKKPHR40 from pro-B cells.The degreeof methylation of the peptide is indicated and the different extents of methylation are accentuated by the shaded envelopes.The singly and doubly methylated peptide ions with m/z1447.8 and 1461.8 were further studied by MS-MS. (e) Portion of the MS-MS spectrum of the m/z 1447.8 singly methylated ion (left panel) and m/z 1461.8 doubly methylated ion(right panel) showing the change in methylation of K27 to K36, K37. The change in methylation is reflected by the change in intensities of the y12-type fragments.The MS-MS spec-trum of the singly methylated ion (left panel) reveals the presence of two ion species with the same mass but having different sites of methylation, whereas the doubly methylatedion (right panel) reveals the presence of three ion species with the same mass but with different sites of methylation.Changes in methylation can be followed by changes in the ratiosof these fragmentation ion intensities. (f) Chromatin immunoprecipitation was done on pro-B cells with anti-pan–methyl-lysine.The coprecipitated DNA was analyzed by PCR withprimers specific for the VHJ558 and VH81X genes.The PCR products were analyzed by Southern blotting with internal probes.
nature immunology • volume 4 no 2 • february 2003 • www.nature.com/natureimmunology
ARTICLES
130
methylation suggested that another lysine residue is controlled by Ezh2in vivo. To reveal the lysine residue within H3 that might be responsiblefor the observed changes in H3 methylation, we employed mass- and tan-dem-mass spectrometry (MS-MS) analysis42,43 of methylation in histoneH3 (Fig. 7d,e). The mass spectra of the proteolytic mixtures obtained bytrypsin digestion of histone H3 derived from the wild-type and Ezh2-deficient pro-B cells were analyzed. MS-MS analysis showed that thismethylation change occurred primarily at K27, where methylation wasattenuated by more than five-fold in Ezh2-deficient pro-B cells comparedto controls (Fig. 7e). Although observed changes in histone methylationappeared to be global, they affected the methylation status of histonesassociated with VHJ558 or VH7183 genes differently (Fig. 7f). In the wild-type pro-B cells, the histone associated with the VH81X locus (VH7183family) was hypomethylated compared to methylated histone linked tothe VHJ558 locus (Fig. 7f). As a consequence, Ezh2 deficiency has a cleareffect on methylation of the VHJ558 but not VH81X locus–associated his-tone. These data explain the selective impact of Ezh2 deficiency onVHJ558 gene rearrangement.
DiscussionEzh2 deficiency leads to diminished generation of pre-B cells and imma-ture B cells in the bone marrow. Defective B cell development cannot berestored by the presence of the wild-type cells in the mixed bone marrowchimeras, thus supporting the B lineage–autonomous nature of theobserved defect. The unaltered in vivo maturation and in vitro activationof B cells that acquire Ezh2 deficiency at post–pro-B cell stages show thatthe requirement for Ezh2 is development stage–specific. It could well bethat Ezh1, which is abundant mostly in mature B cells, compensates forloss of Ezh2.
The cause of impaired Ezh2-deficient B cell development lies in reducedrearrangement of the VHJ558 gene cluster. The fact that the VHJ558 familyof VH genes comprises the largest VH gene family in mouse genome31
explains the reduction in the number of intracellular µ chain–positiveEzh2-deficient pro-B cells.
The absence of Ezh2 reduces both the basal and IL-7–induced histoneH3 lysine methylation. This reduction is mainly due to the diminishedmethylation at lysine 27 (H3-K27). Our results show that similar to theDrosophila protein E(Z)15–17, murine Ezh2 is an HMTase with H3-K27specificity. Furthermore, we have also demonstrated the ability of anextracellular ligand, IL-7, to control chromatin structure via inducible his-tone H3 methylation. Because histone methylation is considered to beirreversible44, even the modest increase in methylation observed in IL-7–stimulated pro-B cells is likely to have a long-lasting impact on thechromatin structure.
The mechanism of histone H3 methylation induction by IL-7 remainselusive. The wild-type–like activation of signal transducers and activatorsof transcription 5 (STAT5), a key signaling component downstream ofthe IL-7R41, makes an involvement of STAT5 in IL-7 induced methyla-tion unlikely. Moreover, unaltered H3 acetylation excludes a possibleindirect effect of Ezh2-deficiency on H3 methylation through changes inhistone acetylation. Overall, these data suggest the existence of STAT5-independent signaling connecting IL-7R and histone H3 methylationthrough Ezh2.
Given the critical role played by IL-7 in the regulation of V(D)Jrearrangement41, in particular of VHJ558 genes39,40, it is plausible that Ezh2-dependent histone H3 methylation provides the mechanism by which IL-7targets the recombination machinery to the Igh locus. Despite the seem-ingly global role of Ezh2 in histone H3 methylation, methylated histoneassociated with VHJ558 or VH7183 genes is not equally affected by Ezh2deficiency. Hence, within the Igh locus, Ezh2 seems to play a selective role
in the regulation of VHJ558 methylation and recombination. Because nei-ther histone acetylation nor VH gene germline transcription are affected byEzh2 deficiency, a mechanism other than transcription must be importantfor V(D)J rearrangement.
The histone code hypothesis45 postulates that covalent modification ofhistones, including acetylation and methylation, promotes binding of spe-cific proteins to chromatin to alter transcription45, the DNA double-strandbreak (DSB) repair46 and DNA excision47. The lack of accumulation ofunrepaired DSBs within the VHJ558 locus in Ezh2-deficient pro-B cellssuggests that DNA cleavage rather than repair are affected by the absenceof Ezh2. Therefore, we speculate that Ezh2-mediated histone H3 methy-lation may facilitate targeting of recombination machinery, includingRAG proteins, which catalyzes the DNA cleavage or is responsible formarking the borders of the DNA excision, similar to the histone methyla-tion–dependent mechanism responsible for programmed DNA elimina-tion that accompanies macronuclear development in Tetrahymena47.Overall, we have identified novel mechanism of regulation of B cell devel-opment through Ezh2-mediated histone H3 methylation. In view of therole of Ezh2 in Igh rearrangement, it remains to be seen whether relatedprocesses such as Ig switch recombination and somatic hypermutation areregulated by histone methylation.
MethodsGeneration of mice with a loxP-flanked Ezh2 allele. A 3.5-kb KpnI fragment of Ezh2 (Fig.2) was inserted into the ClaI site of pKSTKNEOLOXP between the loxP-flanked neor
expression cassette (neor) and gene encoding TK. A 3.5-kb KpnI-BamHI fragment contain-ing the exons encoding SET domain was inserted into the SalI site between neor and the loxPsite. Finally, a 3.5-kb BamHI-MamI fragment was cloned into the NotI site located 3′ of theloxP site. The SacII-linearized DNA of the Ezh2 targeting vector (pKSTKNEOLOXP-Ezh2-SET#15) was transfected by electroporation into E14-1.1 cells48 followed by their selectionin the presence of G418 (300 µg/ml) and gancyclovir (2 µM). The DNA of double-resistantES cells was digested with BamHI and tested for homologous recombination by Southernblot analysis with a 1-kb MamI-EcoRI DNA fragment as a probe (probe A). This probe rec-ognizes 19-kb and 24.5-kb DNA fragments corresponding to the wild-type and targeted loci,respectively. To delete the loxP-flanked neor gene, the ES cell clones carrying the targetedEzh2 allele were transiently transfected with 10 µg of the Cre-recombinase expression vec-tor pIC-Cre22. DNA from neomycin-sensitive clones was analyzed for neor deletion bySouthern blot analysis with probe A and selected ES clones were injected into blastocysts togenerate chimeras and later mice carrying loxP-modified Ezh2. Heterozygous Ezh2fl/+ micewere bred to Mx-Cre mice26 to generate Ezh2fl/+ Mx-Cre mice that were crossed to generateEzh2fl/fl Mx-Cre+/– mice. Mice were genotyped for the presence of the loxP-floxed Ezh2 alleleand Mx-Cre transgene by Southern blotting and PCR. The primers used are listed inSupplementary Table 1 online. The efficiency of inducible Mx-Cre–mediated Ezh2 deletionwas determined by Southern blot analysis of the EcoRI-digested genomic DNA isolated fromvarious lymphoid organs. The results were quantified with the NIH Image 1.62 program. Allmice were bred and maintained under specific pathogen–free conditions at the LaboratoryAnimal Research Center of the Rockefeller University; all mouse protocols were approvedby the Rockefeller University IACUC.
FACS analysis and cell sorting. The preparation of ex vivo–isolated cells for FACScan analy-sis and sorting of the lymphocyte subpopulations on FACSstar was done as described49. Theanalysis of the BrdU-labeled cells and TUNEL assay were performed with a BrdU flow kit (BDPharmingen, San Diego, CA) and a Fluorescein In Situ Cell Death Detection Kit (Roche,Indianapolis, IN) according to the manufacturer’s protocol. Analysis of activated caspases wasperformed with a CaspaTag Fluorescein Caspase (VAD) Activity Kit (Intergen, Norcross, GA)according to the manufacturers’protocol. The antibodies anti-B220 (RA3-6B2), anti-CD43 (S7),anti-IgD (11-26c.2a), anti-Ly51 (BP-1), anti-IgG1 (RB6-8C5), anti-HSA (M1/69), anti-CD127and anti-Ly9.1 were purchased from BD Pharmingen. Anti-IgM was purchased from JacksonImmunoResearch (West Grove, PA). Phycoerythrin (PE)-Cy7 and cychrome-streptavidin wereobtained from Caltag. Anti–IL-7Rα (R7A34), anti–µ chain (M41) and anti–κ chain (R33-18-10)were prepared from the corresponding hybridoma (a gift of K. Rajewsky). For MACS, cells wereincubated with the appropriate magnetic beads (Miltenyi Biotec, Auburn, CA) and purified asdescribed49. The purity of isolated population was controlled by FACS analysis. Purified cells forfurther experiments were at least 95% pure.
RNA isolation, cDNA synthesis and PCR. Total RNA was isolated from 1 × 105 purifiedlymphocyte subpopulations with TRIzol reagent (Gibco-BRL, Gaithersburg, MD) andcDNA was synthesized with the First Strand cDNA Synthesis Kit (Gibco-BRL). PCR reac-tions were performed on a pelitier thermal cycler (PTC 200, MJ Research, Waltham, MA).The primer sequences used in the experiments and references are listed in SupplementaryTable 1 online.
www.nature.com/natureimmunology • february 2003 • volume 4 no 2 • nature immunology 131
EMSA. Cells were washed once with PBS after stimulation and resuspended in lysis buffer (20mM HEPES at pH 7.5, 450 mM NaCl, 0.4 mM EDTA, 0.5 mM dithiothreitol, 25% glycerol, 0.5mM 4-(2-aminoethyl)-benzenesulfonyl fluoride hydrochloride (AEBSF), 10 mM NaF, 1 µg/mlof leupeptin, 2 µg/ml of aprotinin and 5 mM Na3VO4). Samples were subjected to three freeze-thaw cycles. Whole cell extracts were prepared by spinning at 10,000g at 4 °C for 15 min.Extracts (20 µg) in 10 µl of lysis buffer were combined with 11 µl of 2X binding buffer (100mM KCl, 20 mM Tris-HCl, 20 mM HEPES, 1 mM dithiothreitol, 1 mM EDTA and 20% glyc-erol) containing 1 µg of poly (dI)⋅(dC) (Amersham, Piscataway, NJ) and incubated for 20 minon ice. For the supershift, 1 µl (0.2 µg) of STAT5b antiserum (Santa Cruz Biotechnology, SantaCruz, CA) was added to the samples. After a 20-min incubation, 0.4 ng of a radioactive[γ32P]dATP-labeled probe, derived from ovine β-casein (Santa Cruz Biotechnology), was addedand incubated for further 20 min on ice. Samples were resolved on the 5% polyacrylamide nativegels and analyzed after dry gel exposure to x-ray film at –80 °C.
ChIP. Purified pro-B cells (1 × 106) were washed once with PBS and fixed by adding formalde-hyde to a final concentration 0.37% for 10 min at 37 °C. ChIP assays were performed with theChromatin Immunoprecipitation Assay Kit (Upstate Biotechnology, Grand Island, NY) accord-ing to manufacturer’s protocol. Anti-acetyl–histone H3 and H4 (Upstate Biotechnology) andanti-pan–methyl-lysine (Abcom, Cambridge, UK) were used for immunoprecipitation. Theprimers used in the experiments have been described40. The oligoprobes used for the Southernblot analysis are listed in Supplementary Table 1 online.
LM-PCR. DNA from 2 × 105 sorted pro-B cells was isolated with the agarose plug method. Theagarose DNA plugs were subjected to linker ligation for 18 h at 16 °C in 12 µl of ligation buffer(Boehringer Mannheim, Mannheim, Germany) with 48 pmol of linker and 3U of T4 ligase; 1 µlof DNA was used for each PCR reaction. The PCR products were resolved on 1.5% agarose geland transferred to nitrocellulose membrane. The specific bands were visualized by hybridizationwith a radioactive labeled internal probe.
Immunoblot analysis. Immunoblot analysis was performed with standard procedures50 with theuse of nuclear lysate. The nuclei were isolated by incubating cell with nuclei-extraction buffer(320 mM sucrose, 5 mM MgCl2, 10 mM HEPES and 1% Triton X-100 at pH 7.4) on ice for 10min, followed by double washing with wash buffer (320 mM sucrose, 5 mM MgCl2 and 10 mMHEPES). The nuclei were resuspended in sonication buffer (50 mM Tris at pH 8, 500 mM NaCl,1 mM EDTA and 10% glycerol) and subjected to sonication to break the nuclear membrane andextract the nuclear protein. Purified anti–mouse Ezh2 rabbit serum (provided by T. Jenuwein,IMP, Austria), rabbit anti-pan–methyl-lysine (Abcam), rabbit anti-dimethyl–histone H3 lysine 9,rabbit anti-dimethyl–histone H3 lysine 4 (Upstate Biotechnology) and goat anti–histone H3 (N-20) (Santa Cruz Biotechnology) were used in the immunoblotting. Horseradish peroxidase(HRP)–anti-rabbit (Amersham) and HRP–anti-goat (Sigma, St. Louis, MO) were used as sec-ondary antibodies. The signal was detected by the chemiluminescence system (Supersignal,Pierce, Rockford, IL) and quantified with the NIH Image 1.62 program.
Mass spectrometry analysis. Mass spectra of the proteolytic mixtures obtained by trypsindigestion of histone H3 were obtained with an in-house–modified MALDI-QqTOF mass spec-trometer43 with a compact disc (CD) sample stage42. Masses of the tryptic fragments were deter-mined with an accuracy of 10 ppm. After obtaining the tryptic peptide mass map, the CD sam-ple stage was transferred to an in-house–constructed MALDI-ion trap mass spectrometer fordetailed MS-MS analysis of the tryptic peptide ions42.
Web addresses. The DNAPLOT Program Package can be located at http://www.dnaplot.org.
Note: Supplementary information is available on the Nature Immunology website.
Competing interests statementThe authors declare that they have no competing financial interests.
AcknowledgmentsWe thank M. Nussenzweig, K. Rajewsky, C. Schmedt, K. Saijo, I. Mecklenbräuker and D.O′Carroll for discussions.We also thank G. Hannon for critical review of this manuscript.Supported by The Irene Diamond Fund (A.T.), National Institutes of Health grant (A.T.), NIH,RR0086 (B.T.C.) and The Rockefeller University’s Norman and Rosita Winston FellowshipProgram (I.S.).
Received 9 October 2002; accepted 22 November 2002.
2. Mahmoudi,T. & Verrijzer, C.P. Chromatin silencing and activation by Polycomb and trithorax group pro-teins. Oncogene 20, 3055–3066 (2001).
3. Franke, A. et al. Polycomb and polyhomeotic are constituents of a multimeric protein complex in chro-matin of Drosophila melanogaster. EMBO J. 11, 2941–2950 (1992).
4. van Lohuizen, M. et al. Interaction of mouse polycomb-group (Pc-G) proteins Enx1 and Enx2 with Eed:indication for separate Pc-G complexes. Mol. Cell. Biol. 18, 3572–3579 (1998).
5. Sewalt, R.G. et al. Characterization of interactions between the mammalian polycomb-group proteinsEnx1/Ezh2 and EED suggests the existence of different mammalian polycomb-group protein complexes.
Mol. Cell. Biol. 18, 3586–3595 (1998).6. Shao, Z. et al. Stabilization of chromatin structure by PRC1, a Polycomb complex. Cell 98, 37–46 (1999).7. Gould, A. Functions of mammalian Polycomb group and trithorax group related genes. Curr. Opin. Genet.
Dev. 7, 488–494 (1997).8. van der Lugt, N.M. et al. Posterior transformation, neurological abnormalities, and severe hematopoietic
defects in mice with a targeted deletion of the bmi-1 proto-oncogene. Genes Dev. 8, 757–769 (1994).9. Katoh-Fukui,Y. et al. Male-to-female sex reversal in M33 mutant mice. Nature 393, 688–692 (1998).10. Akasaka,T. et al. The role of mel-18, a mammalian Polycomb group gene, during IL-7-dependent prolifera-
tion of lymphocyte precursors. Immunity 7, 135–146 (1997).11. Tokimasa, S. et al. Lack of the Polycomb-group gene rae28 causes maturation arrest at the early B-cell
developmental stage. Exp. Hematol. 29, 93–103 (2001).12. Core, N. et al. Altered cellular proliferation and mesoderm patterning in Polycomb-M33-deficient mice.
Development 124, 721–729 (1997).13. Hobert, O., Jallal, B. & Ullrich,A. Interaction of Vav with ENX-1, a putative transcriptional regulator of
homeobox gene expression. Mol. Cell. Biol. 16, 3066–3073 (1996).14. Hobert, O., Sures, I., Ciossek,T., Fuchs, M. & Ullrich,A. Isolation and developmental expression analysis of
Enx-1, a novel mouse Polycomb group gene. Mech. Dev. 55, 171–184 (1996).15. Czermin, B. et al. Drosophila enhancer of zeste/ESC complexes have a histone H3 methyltransferase
activity that marks chromosomal polycomb Sites. Cell 111, 185–196 (2002).16. Muller, J. et al. Histone methyltransferase activity of a Drosophila polycomb group repressor complex. Cell
111, 197–208 (2002).17. Cao, R. et al. Role of histone H3 lysine 27 methylation in Polycomb-group silencing. Science 298,
1039–1043 (2002).18. Kouzarides,T. Histone methylation in transcriptional control. Curr. Opin. Genet. Dev. 12, 198–209 (2002).19. Raaphorst, F.M. et al. Coexpression of BMI-1 and Ezh2 polycomb group genes in Reed-Sternberg cells of
Hodgkin’s disease. Am. J. Pathol. 157, 709–715 (2000).20. Fukuyama,T. et al. Proliferative involvement of ENX-1, a putative human polycomb group gene, in
haematopoietic cells. Br. J. Haematol. 108, 842–847 (2000).21. O’Carroll, D. et al. The polycomb-group gene Ezh2 is required for early mouse development. Mol. Cell. Biol.
21, 4330–4336 (2001).22. Gu, H., Zou,Y.R. & Rajewsky, K. Independent control of immunoglobulin switch recombination at individ-
ual switch regions evidenced through Cre-loxP-mediated gene targeting. Cell 73, 1155–1164 (1993).23. Laible, G. et al. Mammalian homologues of the Polycomb-group gene Enhancer of zeste mediate gene
silencing in Drosophila heterochromatin and at S. cerevisiae telomeres. EMBO J. 16, 3219–3232 (1997).24. Tschiersch, B. et al. The protein encoded by the Drosophila position-effect variegation suppressor gene
Su(var)3-9 combines domains of antagonistic regulators of homeotic gene complexes. EMBO J. 13,3822–3831 (1994).
25. Carrington, E.A. & Jones, R.S.The Drosophila Enhancer of zeste gene encodes a chromosomal protein:examination of wild-type and mutant protein distribution. Development 122, 4073–4083 (1996).
26. Kühn, R., Schwenk, F.,Aguet, M. & Rajewsky, K. Inducible gene targeting in mice. Science 269,1427–1429 (1995).
27. Baird,A.M., Gerstein, R.M. & Berg, L.J.The role of cytokine receptor signaling in lymphocyte development.Curr. Opin. Immunol. 11, 157–166 (1999).
28. Kitamura, D., Roes, J., Kühn, R. & Rajewsky, K. A B cell-deficient mouse by targeted disruption of themembrane exon of the immunoglobulin mu chain gene. Nature 350, 423–426 (1991).
29. Sonoda, E. et al. B cell development under the condition of allelic inclusion. Immunity 6, 225–233 (1997).30. Rickert, R.C., Roes, J. & Rajewsky, K. B lymphocyte-specific, Cre-mediated mutagenesis in mice. Nucleic
Acids Res. 25, 1317–1318 (1997).31. Wu, G.E. & Paige, C.J.VH gene family utilization in colonies derived from B and pre-B cells detected by
the RNA colony blot assay. EMBO J. 5, 3475–3481 (1986).32. Connor, A.M. et al. Mouse VH7183 recombination signal sequences mediate recombination more fre-
quently than those of VHJ558. J. Immunol. 155, 5268–5272 (1995).33. Ehlich, A. et al. Immunoglobulin heavy and light chain genes rearrange independently at early stages of B
cell development. Cell 72, 695–704 (1993).34. Li, S. & Wilkinson, M.F. Nonsense surveillance in lymphocytes? Immunity 8, 135–141 (1998).35. Schlissel, M.S. & Stanhope-Baker, P. Accessibility and the developmental regulation of V(D)J recombination.
Semin. Immunol. 9, 161–170 (1997).36. Grawunder, U. & Harfst, E. How to make ends meet in V(D)J recombination. Curr. Opin. Immunol. 13,
VH genes occurs in the newly generated repertoire of neonatal and adult mice. J. Exp. Med. 171,843–859 (1990).
38. Lin,W.C. & Desiderio, S.V(D)J recombination and the cell cycle. Immunol.Today 16, 279–289 (1995).39. Corcoran,A.E., Riddell,A., Krooshoop, D. & Venkitaraman,A.R. Impaired immunoglobulin gene rearrange-
ment in mice lacking the IL-7 receptor. Nature 391, 904–907 (1998).40. Chowdhury, D. & Sen, R. Stepwise activation of the immunoglobulin mu heavy chain gene locus. EMBO J.
20, 6394–6403 (2001).41. Hofmeister, R. et al. Interleukin-7: physiological roles and mechanisms of action. Cytokine Growth Factor Rev.
10, 41–60 (1999).42. Krutchinsky,A.N., Kalkum, M. & Chait, B.T. Automatic identification of proteins with a MALDI-quadru-
pole ion trap mass spectrometer. Anal. Chem. 73, 5066–5077 (2001).43. Krutchinsky,A.N., Zhang,W. & Chait, B.T. Rapidly switchable matrix-assisted laser desorption/ionization
and electrospray quadrupole-time-of-flight mass spectrometry for protein identification. J.Am. Soc. MassSpectrom. 11, 493–504 (2000).
44. Bannister,A.J., Schneider, R. & Kouzarides,T. Histone methylation: dynamic or static? Cell 109,801–806 (2002).
45. Strahl, B.D. & Allis, C.D.The language of covalent histone modifications. Nature 403, 41–45 (2000).46. Bird,A.W. et al. Acetylation of histone H4 by Esa1 is required for DNA double-strand break repair. Nature
419, 411–415 (2002).47. Taverna, S.D., Coyne, R.S. & Allis, C.D. Methylation of histone H3 at lysine 9 targets programmed DNA
elimination in tetrahymena. Cell 110, 701–711 (2002).48. Torres, R.M. & Kühn, R. Cre/loxP recombination system and gene targeting. Meth. Mol. Biol. 180,
175–204 (2002).49. Mecklenbräuker, I., Saijo, K., Zheng, N.Y., Leitges, M. & Tarakhovsky,A. Protein kinase Cδ controls self-anti-
